Current and New Ways to Measure Shared Mobility for Australian Cities – the Case of South East Queensland’

How much transport is shared? Current and new ways to measure shared mobility for Australian cities – the case of South East Queensland’

Abraham Leung Cities Research Institute, Griffith University [email protected]

State of Australian Cities Conference and PhD Symposium 30th November – 5th December, 2019 Perth, Western Australia www.soac2019.com.au SOAC 2019 How much transport is shared? Current and new ways to measure shared mobility for Australian cities – the case of South East Queensland

Abraham Leung Cities Research Institute, Griffith University

Abstract: New forms of shared mobility, such as ride-hailing, car-sharing, bike- and scooter-sharing, are all touted as promising means to address car dependence and mass car ownership. Most Australian cities and their policymakers are devising policies to facilitate the entry of these new modes. Many have, however, disputed whether many of these new disruptive transport innovations are indeed sharing and several knowledge gaps remain about shared mobility. Transport policy is overwhelmingly focused on increasing transport supply by building more infrastructure. Demand-side responses such as increase “sharing” of modes receive comparatively less research and policy attention. A key challenge is the limited understanding of “how much transport is shared” and “how to make sharing easy”. Conventional household travel surveys and other forms of data collection are ‘blind’ to whether travellers are either sharing vehicles or sharing rides. In this paper, we explore fundamental questions about sharing in Australian cities – How should we define and measure sharing? How much transport is shared? How could we better measure and capture it? Most importantly, how much of the currently non-shared travel is amenable to sharing? The methods involve in-depth review and analysis of South East Queensland Travel Survey in 2017-18. The results provide preliminary analysis on the state of shared mobility from existing data, and propose improved protocols and new data sources. Urban policymakers should embrace and encourage shared mobility but not ignoring the potential pitfalls.

Keywords: shared mobility, ride hailing, ride sharing, South East Queensland

Introduction Sharing is today a zeitgeist in the transportation field with disruptive changes are happening at a rapid pace in transport. Shared mobility is being boasted as an innovative way to resolve many urban transport problems such as congestion and its associated costs such as pollution (Nykvist and Whitmarsh 2008). Innovative sustainable mobility practices, such as car and bike/scooter sharing are rapidly growing, which are highly applicable for Australian cities - being highly car dependent with a majority of trips are done by private vehicles with high proportion of single or low occupancy. Widespread subscription to collectivised fleets of on-demand vehicles linked to other forms of public transport, public bike hire schemes, and other shared platforms set to create a new “ecosystem” that transport planners call Mobility-as-a-Service (MaaS) – one may order and pay for all trips by smartphone. Understanding how, when and why Australians will share their car trips and vehicles, now and in the future, is fundamental to helping transition our cities to improved transport arrangements. Current ways to collect transport behaviour information are at the risk of being outdated as new forms of sharing are increasingly difficult to classify or define.

The paper first presents the latest definitional and conceptual understanding of shared travel, demonstrating its versatility and varied nature – sharing as a multifaceted set of modes that meet the needs of different people at different times. Then, the latest South East Queensland Travel Survey (SEQTS) dataset that incorporated new questions and variables regarding mode and trip-specific shared transport are analysed. The data limitations of SEQTS are highlighted, and recommendations are proposed to help address current data deficiencies and barriers to sharing. The paper concludes by considering how sharing such may usher wider social change and make sharing transport part of life in a sustainable manner.

The state of shared mobility worldwide Shared mobility is an increasingly used term of denotes “transportation services and resources that are shared among users, either concurrently or one after another” (Shared-Use Mobility Center 2019). This encompassed incumbent modes such as public transport; taxis; carsharing; ridesharing; ridehailing; ride- splitting; bike and scooter sharing (increasingly referred as “micromobility” (Dediu 2019)), etc.

How to encourage travellers to share transport with other passengers is now attracting significant attention from transport and urban policy practitioners and researchers. With the advancement in smartphones and information technology, there is now a myriad of ways to share travel: Uber claims that in cities where 2

SOAC 2019

UberPOOL is available, over 20% of trips are “ridesplit” - where shared-trip payments involving multiple passengers, with more than one start or end point (i.e. ‘true’ ride-sharing). There are also non-profit services found in Sweden where for-profit sharing is outlawed. The non-profit service Skjutsgruppen facilitates ridesharing of car and boat rides. Sunfleet offers car-sharing services, whilst schemes like Car2Go have recently failed there. Internationally, new peer-to-peer (P2P) car sharing services like CarNextDoor allow users to share their private vehicle as on-demand hire cars to others, which is essentially a vehicle version of Airb2b. In Europe and North America (to be launched in Australia), Car2Go, a free-floating car sharing system is revolutionising the car rental industry. It bundles car sharing with parking results in better convenience and higher utilisation potential (Münzel et al. 2019). Crowdshipping is also gaining popularity as a cheaper way to collectivise shipping and unused freight capacity (Punel et al. 2018). Google’s Waymo has suggested the future of autonomous cars is potentially the abandonment of mass private car ownership and, instead, subscription to collectivised fleets of on-demand vehicles.

These new forms of transport sharing may connect to various public transport, public bike hire schemes, and other shared platforms to create an integrated MaaS system that multimodality might replace traditional car ownership that encourages driving. But there is a major stumbling block to this vision: government policy makers know little about the propensity for sharing of car transport and their wider impacts due to the lack of useful data. Informal ways sharing, such as offering a lift to someone are often not recorded in travel surveys. True rate of sharing is unknown. Understanding how, when and why urban residents will share their vehicle seating/space and vehicles itself, now and in the future, is fundamental to helping transition our cities to improved transport arrangements.

What is shared mobility? Thanks to technological advances in internet and handheld mobile dervices, increasing numbers of consumers are opting for temporarily accessing or sharing products and services instead of buying or owning them (Standing et al. 2018). This is set to replace ownership-based consumption. Using the collaborative sharing framework by Scaraboto (2015), sharing transportation can be conducted in the following ways:

1. For monetary profit: • Purchasing a service – pay for someone for the temporary use of a vehicle (ride-hailing/sharing). • Renting – a vehicle can be rented rather than purchased (car-rental, bike-sharing) • Subscribing – people can subscribe for the access of the vehicle (car clubs, subscription-based bike- sharing).

2. Not-for-profit: • Donating – people can give free rides in their vehicle (car-pooling - donating time, donated bicycles) • Lending – a vehicle can be borrowed or loaned (car/bike sharing). • Exchanging – non-monetary exchange of goods or service (such as bartering).

Sharing in transportation is becoming a profitable business, which is now expanding into a wide range of shared mobility services (such as Uber, Lime, Mobike). Shasheen et al. (2016) summarised the common forms of shared mobility service models (Figure 1).

In summary, shared mobility may differ in the following ways:

• Mode (car, bicycle, e-scooter, public transport);

• What is being transported (persons or freight);

• The ability of control by the user (“ridesharing” versus “carsharing” – the later allows full control of vehicle akin to car rentals);

• Number of seats in a vehicle (ranging from single-seaters (single-seated motorcycles, which are unable to share rides) to high occupancy public transport);

• Ownership of vehicle (fleet-based or peer-to-peer sharing);

SOAC 2019

• How demand is matched (e.g. the level of “self-service”); and

• Level of technology (phone-booking or now increasingly done by mobile phone apps and internet, with algorithms for big data analysis and artificial intelligence in predicting demand).

Figure 1: Existing and Emerging models for shared mobility services

Source: Shasheen et al. (2016)

These considerations dictate the lens required for analysing shared mobility. The distinction between “vehicle-sharing” and “ride (occupancy) sharing needs to be made (Currie 2018). Vehicle (or car) share refers to sharing the ownership of a vehicle. Rides (or trip) share refers to sharing the occupancy of a vehicle while travelling. The impact of different forms of sharing varies. Generally speaking, if more occupancy is shared, it could cause fewer externalities, such as road congestion or environmental impact. Calculating region-wide vehicle occupancy is a more difficult task as travel surveys typically track private vehicle use only.

Perhaps due to data availability, a significant focus of research has been on the sharing private vehicles as a mode. If sharing reduces car ownership and habitual driving, it may address car dependence and congestion. Early research has conceived of ridesharing and carsharing as static, pre-organised sharing (Horowitz and Sheth 1976, Teal 1987, Giuliano 1992, Ferguson 1997, pp. 1970–1990). Recent growth in information technology allowed “peer-to-peer” and more dynamic sharing possibilities (Kwan 2007, Buliung et al. 2010), with a greater emphasis been put on workplace-based schemes. Only until recently, research expanded to app-based sharing and the potential of autonomous vehicles (Furuhata et al. 2013, Cetin 2017, 4

SOAC 2019

Sprei 2018). For the case of car sharing, the only known studies of in the Australian context are general transport studies (Hensher 1998, Mees et al. 2008) and some cases studies of carsharing policy (Dowling and Kent 2015). Travel behaviour changes due to ridehailing remain limited, perhaps due to its relative nascent development. However there are signs of reduction in personal driving and might replace local bus services and are serve as a connector to longer-distanced modes (Feigon and Murphy 2016, Clewlow and Mishra 2017, Circella and Alemi 2018, Henao and Marshall 2018). Other benefits such as reducing traffic fatalities and driving under the influence were also noted (Brazil and Kirk 2016, Young and Farber 2019). Environmental awareness, the “peak car”, economic recession and changing demographics might explain the recent declining preference to own a car in developed countries (Hopkins 2016). Nevertheless, detailed and up-to-date car or ridesharing user analysis remains sparse and limited in Australian scholarship, this may attribute to lack of reliable datasets on a city/region-wide basis.

Data available – How much can we know about shared mobility? In this paper, I focus on mode-specific shared mobility based on “trips” in a regionally aggregated view. Certain modes must be “shared”, such as “being a passenger”, be it private or public transport. While some modes could be shared or not, such as “being a driver”. While this view is somewhat simplistic, it is a rough measure of overall sharing at a broader scale, given the fact that overall journey to work car occupancy is only slightly over one during peak hour in most Australian capital cities. Vehicle occupancy data is limited at regional scale - accurate occupancy data can be obtained from manual observations of road traffic but are localised in nature. Passenger data from shared mobility operators (such as Uber) only limits to their own business - only summarised statistics are available to the public at best. Government conducted transport surveys, on the other hand, are publicly available and covers larger areas. For national level, the quinquennial nationwide census conducted by the Australian Bureau of Statistics (ABS) captures a “snapshot” of the entire population regarding journey to work (JTW) alongside with many socio-demographic data. While this offers the most extensive coverage of the whole population, it only captures work-related trips, with education and shopping trips ignored. This shortcoming can be addressed by State/Territory level household travel surveys (HTS), which sample around 5-10% of the population. HTS captures the number of trips made in a period (usually a week) using travel diaries, which includes more detailed information, including non-work trips.

Shared mobility has not been satisfactorily captured in both surveys but it is improving. The latest ABS 2016 Census did not include ridehailing modes (such as Uber) as a default option in the JTW questions. The most recent South East Queensland Travel Survey (SEQTS) conducted in 2017-8 by the Queensland Department of Transport and Main Roads (TMR) introduced new questions regarding ridehailing (Uber specifically) as it became legalised in Queensland. This study examines the result of SEQTS pertaining shared mobility and evaluate whether the data are sufficient for academics and practitioners, and also the implications to urban planning and policy.

The Southeast Queensland Travel Survey (SEQTS) The SEQTS 2017-2018 obtained 7-day travel information of 13,726 respondents living in 5,460 households at South East Queensland, recording 40,368 valid trips. The figures were weight-adjusted to represent the entire population using the included personal weights – the resultant data represents in total 3,278,776 persons, 1,116,623 households and 9,625,380 trips, serving as the basis of analysis in this paper.

Travel diaries for a week were completed for all members of the household (over 5yo), capturing information on all trip stages for each trip. The “main mode” in SEQTS refers to the highest-order mode used on any trip stage in an overall multi-modal trip based on (1) the longest-time; and (2) the priority mode based on hierarchical order of collective transport modes, private vehicle travel, cycling, and lastly walking. The key results regarding shared mobility are presented below, showcasing recent trends and also for evaluating the usefulness of SEQTS in understanding shared transport.

Overall travel and sharing behaviour in the survey week Table 1 shows the main mode share of trips by their purpose. Overall, the most popular method of travel (most trips made) in SEQ is being a “car driver” (56.23%). The mode “car passenger” is a crude way to infer the level of private carpooling – in which a total 24.98% of all trips are being driven by someone else. Public transport (including train, public buses, ferry and light rail) as a form of sharing only account for 4.91% of all trips. School transport (including chartered or numbered buses) account for 0.89% of all trips. Taxi is an incumbent transport sharing service characterised by a “for-hire” driver. Perhaps due to the rise of 5

SOAC 2019 ridehailing, now it only accounts for 0.13% of all trips. Whereas ridehailing now enjoys an overall mode share of 0.44%. Active travel, including walking (8.98%) and cycling (1.49%), together comprise of 10.47 of all trips. Other modes (1.95%) include minor modes such as charter buses, trucks, motorcycles and mobility scooters.

The dominant modes that require sharing occupancy account for 31.35% of all trips. A high proportion of these shared trips are for “accompany someone” (84.41%) and for education (79.3%). It is evident that most people do not wish, or not able to share their travel for direct work commute (18.72%) and other work-related (11.21%) trips. Table 2 is similar to Table 1 but sorted by the proportion of trip purposes of the same mode, reflecting the overall of purpose of a particular mode. For better legibility, these values are colour-coded by their strength in percentage values (heatmapping).

Table 1: Mode share of trips by purpose

Trip purpose (%) Main mode Accompany Direct work Other work Personal Education Recreation Shopping Social All trips someone Commute related business Public transport N=472,476 1.05 12.18 3.22 11.24 4.63 1.38 2.51 2.41 4.91 School transport N=85,620 0.04 0.03 0.20 7.28 0.12 0.33 0.02 0.00 0.89 Taxi N=12,834 0.06 0.11 0.48 0.00 0.39 0.04 0.18 0.26 0.13 Ridehailing N=55,299 0.12 0.41 0.58 0.02 0.39 0.88 0.45 2.80 0.44 Car passenger N=2,404,392 82.81 5.72 5.68 56.37 21.00 20.26 21.16 22.18 24.98 All shared N=3,017,787 84.09 18.45 10.16 74.91 26.53 22.89 24.32 27.66 31.35 Active travel N=1,007,569 11.79 3.29 9.88 14.92 6.98 30.57 9.39 6.79 10.47 Car driver N=5,412,026 3.75 75.01 71.83 9.42 64.57 44.60 65.41 63.24 56.23 Other modes N=260,784 0.37 3.25 8.13 0.76 1.92 1.94 0.88 2.31 1.95 Total (overall) N=9,625,380 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

Table 2: Purpose of trip by modes

Accompany Direct work Other work Personal Main mode Education Recreation Shopping Social All trips someone Commute related business Public transport N=472,476 2.01 47.27 4.03 25.05 5.83 3.08 8.62 1.79 100.00 School transport N=85,620 0.46 0.58 1.35 89.56 0.86 4.11 0.42 0.00 100.00 Taxi N=12,834 4.06 16.28 22.18 0.00 18.07 3.33 22.35 7.19 100.00 Ridehailing N=55,299 2.58 17.66 8.14 0.45 5.40 21.93 17.12 23.10 100.00 Car passenger N=2,404,392 31.06 4.37 1.40 24.69 5.19 8.91 14.27 3.23 100.00 All shared N=3,017,787 25.13 11.21 1.99 26.14 5.23 8.02 13.07 3.21 100.00 Active travel N=1,007,569 10.55 5.99 5.80 15.59 4.12 32.10 15.12 2.36 100.00 Car driver N=5,412,026 0.63 25.42 7.85 1.83 7.10 8.72 19.60 4.09 100.00 Total (overall) N=9,625,380 9.37 19.06 6.15 10.94 6.18 10.99 16.85 3.63 100.00 Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

SOAC 2019

Generally speaking, shared trips are more likely used for education (26.14%) and accompanying someone (25.13). Whereas car driver trips are usually attributed to direct work commute (25.42%) and shopping (19.6%). Active travel trips are typically recreational (32.1%), to a lesser extent, education (15.59%) and shopping (15.12%).

Mode specific wise, the most common trip purpose of public transport is direct work commute (47.27%) followed by education (25.05%). School transport is dominated by education trips (89.56%). A large proportion of car passenger trips are for accompanying someone (31.06%) and education (24.69%) – this also serves a social function as being a passenger utilises the car as a private communal space (albeit in motion). There are some nuanced differences between taxi and ridehailing. Taxis are more likely used for other work-related (not commuting, 22.18%), personal business (18.07%) and direct work commute (16.28%) trips. On the other hand, ridehailing is found popular on social trips (23.1%), recreation (21.93%) trips. Similar to taxi, ridehailing is also commonly used for direct work commute (17.66%) and shopping (17.12%) trips.

The SEQTS data also reveal the time of travel (Figure 2), which is not available in ABS census. Active modes, car, and public/school transport usually occur during usual travel times at daytime (From 7am to 6pm), mirroring the typical work or school commute. School transport unlikely occurs in off-peak hours. Car driving is slightly more likely in off-peak night hours, possibly reflecting less congestion levels in these hours. Perhaps due to the lack of alternatives, taxi and ridehailing tend to fill the transport demand during off-peak hours. But both demand responsive modes covers different non-peak times: taxi trips are most likely (66.11%) to occur during off-peak hours of the day (9am to 4pm), but ridehailing trips are much more likely (58.94%) to happen during off-peak hours at night (6pm to 7am). Household travel surveys also capture useful information of travel time and distance. Some interesting patterns are also observed, as presented in Figures 3, 4 and 5.

Figure 2: Starting time of the trip by modes (%) 70

Proportionofstartingby trips (%) time 10

0 Public School Taxi Ridehailing Car Active modes Car driver Overall Transport Transport passenger

Shared modes Non-shared modes

AM Peak Off-Peak Day PM Peak Off-Peak Night (7am to 9am) (9am to 4pm) (4pm to 6pm) (6pm to 7am)

Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

SOAC 2019

Figure 3: Travel time of selected modes

mean Train Public bus Ferry Light rail Shared modes School Bus Taxi Ridehail Car passenger Walking Non-shared Bicycle modes Car driver All modes

0 20 40 60 80 100 120 140 160 180 200 Travel time (minutes)

Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads Note: the box whisker chart shows the 2%, 25%, 50%(median), 75% and 98% percentile. The red circle denotes the mean value.

Figure 4: Travel distance of selected modes

mean Train Public bus Ferry Light rail Shared modes School Bus Taxi Ridehail Car passenger Walking Non-shared Bicycle modes Car driver All modes

0 20 40 60 80 100 120 Travel distance (km)

Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

SOAC 2019

Note: the box whisker chart shows the 2%, 25%, 50%(median), 75% and 98% percentile. The red circle denotes the mean value.

Figure 5: Mean travel speed of selected modes

5 Mean travelspeed (km/h)

0 Train Public bus Ferry Light rail School Taxi Ridehail Car Walking Bicycle Car driver All modes Transport passenger

Shared Non-shared modes modes

Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads (Note: derived from mean travel time and distance)

Travel time and distance both strongly related to the speed of the mode, as the “Marchetti’s (1994) constant” suggests the average total time spent by a person to commute is approximately one hour (both ways). Obviously, active travel trips are limited by their slow speed and short distance. Motorised modes can go further and faster. For shared modes, high occupancy modes such as public transport tend to follow pre- defined route, stops more and with more detours - this results in a higher than mean travel time or distance. Rail and ferry trips tend to have longer travel duration and distance, while bus trips tend to be shorter.

However, public transport remains competitive for certain long distance CBD-bound trips (such as from the Gold Coast) over car-based modes due to congestion and high parking costs. For shared car trips, car passenger, taxi and ridehailing trips exhibit similar patterns in terms of travel time, distance and speed. Some subtle changes can be detected – taxi trips tend to be shorter as the price is higher, and is more sensitive to distance.

The time and distance travelled also reflects on speed of travel. Car-based modes are usually the fastest, followed by public transport. Buses tend to stop more than ferry or trains. Active travel, in particular walking, is the slowest of all modes considered. Taxis are slower than ridehailing, probably reflecting their usage in day time with more traffic. Both are closer than private car use – a possible reason is that hired drivers might drive more cautiously.

Long term patterns of rideshare and taxi use Apart from the above results during the weekly trip diary data, the SEQTS also included a new question about the long term patterns of both rideshare and taxi use, which allows the analysis of habitual use of hired driver use in SEQ. The questions asked how often taxi or ridehailing trips were made within 30 days. Table 3 summarises the results of this data showing some distinctive pattern of taxi and ridehailing trips.

A significant portion of respondents (76.56%) reported that not making a taxi trip within 30 days, but only (2.41%) indicated such for ridehailing trips. Such finding might reflect the cannibalisation of for-hire vehicles Ridehail benefited from the rapid rise of TNCs (such as Uber, Didi or Ola) in the SEQ market. Significantly, more respondents have made a ridehailing trip for 1-3 times within 30 days (51.51%). Similar to weekly trip patterns, most of the regular trips were for social and entertainment purposes.

SOAC 2019

Table 3: Number of for-hire vehicle trips made within 30 days by trip purpose

Trip purposes (multiple choice) Social/ Medical Education Shopping Work Other TOTAL TAXI trips in made last 30 days Entertainment 0, made but not within 30 days 132,263 8,064 7,172 9,646 40,979 23,065 221,190 1-3 30,972 1,253 510 1,233 9,252 5,158 48,378 4-6 6,466 285 530 0 4,756 641 12,679 7-9 447 0 0 375 761 245 1,828 10-19 1,012 174 89 53 726 296 2,351 20+ 0 0 0 651 1,851 0 2,502 TOTAL 171,160 9,777 8,301 11,958 58,325 29,405 288,927 Social/ Medical Education Shopping Work Other TOTAL RIDEHAIL trips made last 30 days Entertainment 0, made but not within 30 days 3,158 208 163 167 2,043 1,459 7,198 1-3 101,772 3,822 3,124 4,611 22,745 17,852 153,926 4-6 46,175 2,563 1,877 1,600 15,338 6,029 73,582 7-9 11,861 743 1,288 256 4,944 697 19,788 10-19 11,627 2,071 1,585 1,629 7,602 2,396 26,911 20+ 3,170 1,339 892 3,847 6,817 1,351 17,416 TOTAL 177,763 10,745 8,930 12,110 59,490 29,783 298,821 Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

The long term usage levels can also be cross-examined with demographic variables such as age, car licence possession rates, employment and life stages (Table 4). Frequent taxi or ridehailing users are associated with full-time work. This reflected the recent trends of ridehailing becoming more dominant that taxi, yet our analysis is not longitudinal (Conway et al. 2018). Taxi users are generally characterised by higher age, less likely to possess car licence, not in part-time work, and are more likely to be retirees whilst ridehailing users are mostly the opposite. The socio-demographic patterns are similar – with younger individuals are found more receptive to ridehailing.

Table 4: Selected demographic variables percentages for hire trips made within 30 days

Selected demographic variables % with car % work full- % work part- % at uni. full- Mean Age $ at school % retired TAXI trips in made last 30 days licence time time time 0, made but not within 30 days 39.72 78.59 38.56 13.26 5.31 16.01 14.23 1-3 45.72 89.86 58.80 14.10 4.52 2.00 14.86 4-6 45.65 86.08 61.79 9.71 3.24 1.61 14.40 7-9 53.10 66.12 56.29 0.00 8.37 0.00 32.30 10-19 48.85 70.42 57.63 4.74 0.00 1.96 19.60 20+ 38.26 82.83 83.12 0.00 0.00 3.55 2.72 RIDEHAIL trips made last 30 % with car % work full- % work part- % at uni. full- Mean Age $ at school % retired days licence time time time 0, made but not within 30 days 40.92 77.00 37.12 12.44 4.22 16.85 16.08 1-3 36.45 92.55 57.56 19.14 9.02 4.43 4.09 4-6 35.66 92.98 61.19 14.06 13.93 2.30 3.24 7-9 35.37 96.89 55.93 6.85 26.18 0.00 5.78 10-19 35.70 93.75 53.70 15.82 13.86 1.25 6.36 20+ 40.75 90.56 62.08 15.09 0.00 6.40 7.64 OVERALL 37.19 72.88 36.35 12.62 4.72 15.24 13.39 Source: 2017-18 South East Queensland Travel Survey (SEQTS), Queensland Department of Transport and Main Roads

SOAC 2019 How could we better measure shared mobility? While SEQTS is quite apt in revealing the key patterns of shared mobility, there are a number of limitations of this dataset. We propose several recommendations so as to improve the overall understanding of shared mobility in future, and also ways to improve sharing uptake.

Better ways to collect shared mobility information in household travel surveys While household travel surveys obtain reasonable sample sizes that can help calibrate travel models and understand key trends in travel behaviour, the data is usually cross-sectional (not longitudinal) and there are changes in methodology between rounds may reduce comparability. Furthermore, conventional travel surveys/diaries require respondents to recall their previous travel (self-reported data) causing accuracy problems. Tracking using global navigation satellite systems (GNSS), such as Global Positioning System (GPS) can record positions of travellers within 10m. Combined with smartphones, researchers could track increasingly spatially detailed data about travel (Hardy et al. 2017, Korpilo et al. 2017). Smartphone data collection has increasing capabilities for interactive questions to be “popped-up” during critical moments. For instance, when a traveller arrives at a point-of-interest that is geo-fenced (Lovett and Peres 2018, Kaufman et al. 2019). Passively collected mobile phone data contains large samples of users and are increasingly adopted in transport research – for example, the Strava app has been used in cycling research in Queensland (Heesch et al. 2016, Sun and Mobasheri 2017). While passive tracking may track a large number of travellers, it is unable to infer socio-demographic data. There remains a strong need to survey respondents actively. If smartphone-based tracking apps can be developed with a long-term panel for tracking the same population over time regarding shared mobility, this could address a significant stumbling block of research in this kind.

Addressing sample size issues Recent ridehailing studies varied in terms of sample size. Ranging from small to large: dedicated surveys (Rayle et al. 2016, Saravanan Sundara Sakaran et al. 2018), household travel surveys with ridehailing as mode option (Young and Farber 2019) or operational data from shared mobility providers (Yim 2019). Another aspect is the quality of data – perhaps only the operator would be able to collect fine-scaled spatial and temporal data which could pin-point and geolocate travel patterns. SEQTS data are not as sophisticated in terms of shared mobility data collection yet – it only offers self-reported travel dairy data collected the origin and destination location of each trip. While such analysis would be useful for main modes such as car passengers or public transport, however, as only a small number of taxi/ridehailing trips were collected (only around raw 200 trip samples), such is unlikely to yield detailed analysis due to concerns of small sample error effects. The data presented earlier in this paper are only useful by aggregation into larger groups and combined with the use of population weightings are used to approximate the wider population.

Making use of operational datasets from service providers Although vehicle trip data in household travel travels, including car passenger trips, occupancy of trips can only be indirectly inferred. Carpooling patterns can be challenging to elicit as SEQTS does not have questions regarding the arrangement of formal/informal car polling. As such, researchers are only able study typical “hire-booking” modes (taxi or ridehailing services). This is particularly challenging for studying bike sharing (such as using Brisbane’s CityCycle), which in SEQTS it would be assumed as “cycling” in the mode share selection by respondents. Recently emerging sharing modes such as e-scooter sharing would also be difficult to obtain as there is no such option in the surveys. For these modes, relying on shared mobility operators to provide such information to the transport authorities is required, but data collection and sharing standards are still being developed.

While some transport network companies (such as Uber or Lime) were obliged to provide data to authorities, these are not available in the public domain for researchers to study. Furthermore, these operational use data are likely not conducted alongside with household surveys, which may prove difficult to compare with SEQTS. Fusing multiple datasets is touted as a promising way to address data limitations in understanding shared mobility (Dias et al. 2019). Better information could create important baselines for future transport research. Future travel surveys should also enlist cooperation from shared mobility providers.

Data standardisation efforts Many forms of shared travel are still informal and invisible, for instance, non-profit peer-to-peer transport sharing (Golightly et al. 2019). Future travel survey should incorporate questions regarding sharing behaviour. For formal taxi or ridehailing services, there is a need from transport authorities to devise data 11

SOAC 2019 collection protocols to collect essential transport data. There increasingly popularity of online platforms may allow more such kind of data to be collected. This is, however, not a legal requirement. Legislation that regulates ridehailing (and even bike/scooter sharing) should also consider data collection in mind. Following the footsteps of public transport data harmonisation (e.g. General Transit Feed Specification (GTFS) by Google), similar efforts are underway for bike or scooter sharing and usage, such as mobility data specification (MDS) and General Bikeshare Feed Specification (GBFS). Ridehailing and other emerging form of sharing warrant data standardisation as well, an example is New York’s Taxi and Limousine Commission’s Trip Record Data. These efforts are worryingly lacking in Australia.

While improving the understanding of shared mobility is essential, policy measures are also necessary for making sharing “irresistible” – both in terms of vehicle/ride-sharing. However, safeguards need to be in place to ensure sharing can be promoted while being sustainable. The following aspects are explored here.

Acceptability and ease of use Sharing is also a multimodality problem which the “transfer penalty” between modes could be an inhibiting barrier. Current transport sharing solutions are typically used in multimodal transport (e.g. transferring from one mode to another) and not necessarily for the whole trip ('door-to-door'), which tends to be car-based but at a greater cost. One is more willing to share vehicles or rides when multiple accessible alternatives are in place, and the car is not the only feasible mode. It would be a missed opportunity if ridehailing services are allowed to cannibalise public transport, and it has already happened in outer urban North America (Smith 2017). Instead, all forms of mode needs to work together as MAAS - integrated ticketing in Helsinki paved the way for using public transport, bicycle sharing, carsharing and also ridehailing in mobile phone apps (such as Whim). The next-generation Smart Ticketing Project by Translink (2019) may help to unlock the payment and platform barrier to seamless travel using an account-based ticketing system. Last but not least, an attractive urban environment that is easy, safe and comfortable to walk and cycle is also an often forgotten incentive that induces multimodal transfer and reduces the need for environmentally damaging door-to-door modes (Circella et al. 2019).

Cost and incentives If there is an integrated platform to use and share transport, multiple transport services could be bundled together and to provide incentives for high volume users. Various TNCs are already providing bundles for ridehailing, courier, food delivery and even loyalty bonuses, such as Grab in South East Asia. The business model of Netflix, which offers periodical subscriptions with varying price levels can be applied in transport, too (Kaufman and Leung 2019). Such schemes reward users to consume more develop habitual use. Linking subscriptions with a new ticketing system have been one of the success factors of Whim in Helsinki, Finland.

Governance and sustainability Despite the promising claims of shared mobility that it could reduce car dependence and address externalities, there is a real danger that unsustainable business practices could do more harm to the environment (increased empty ridehailing trips and congestion(Li et al. 2016), clogged streets with unused shared bikes(Gu et al. 2019)) and cause social inequity (exploitation of low-wage labour due to weakened collective bargaining power (Saadah et al. 2017, De Paula and Zanatta 2018)). Governments need to step up to enact legislation that regulates this new shared economy. Some jurisdictions (such as Vancouver (Mobility Pricing Independent Commission 2018)) are also mooting to use road pricing as a way to tackle with the potential adverse external effects caused by replacement of car ownership with vehicle/ride-sharing.

Concluding Remarks How, and to what extent, improved mobility and accessibility can be offered by shared mobility remains difficult to answer due to lack of understanding. How much of the currently non-shared travel is amenable to sharing remains unknown and needs further research and policy attention. The actual rates of ride/car sharing occurring across transport markets are only limited for prevailing and formal modes. The drivers of and barriers to sharing in contemporary Australian cities remain sketchy and limit our knowledge of likely consumer responses to shared vehicle models under emerging MaaS paradigms. Transport policy and planning responses to ensure sharing must be promoted while minimising the negative impacts (e.g. road space usage, social equity and public liability).

SOAC 2019

This paper outlined current understanding using Queensland’s example of periodical, cross-section surveys (SEQTS 2017-8). While the survey included new variables covering ridehailing – both weekly and monthly usage patterns, limitations remain in capturing the real extent of shared mobility modes, in particular micromobility, ridehailing/pooling and carsharing/pooling. More importantly, there is scant consideration of real-time, geocoding and longitudinal aspect of travel data collection. Much of the sharing has not been captured. While better data collection may help to plan for adapting shared mobility, embracing shared mobility sustainability with better urban policy remains urgently needed.

References Brazil, N. and Kirk, D.S., 2016. Uber and Metropolitan Traffic Fatalities in the United States. American Journal of Epidemiology, 184 (3), 192–198. Buliung, R.N., Soltys, K., Bui, R., Habel, C., and Lanyon, R., 2010. Catching a ride on the information super- highway: toward an understanding of internet-based carpool formation and use. Transportation, 37 (6), 849–873. Cetin, T., 2017. The Rise of Ride Sharing in Urban Transport: Threat or Opportunity? In: H. Yaghoubi, ed. Urban Transport Systems. InTech. Circella, G. and Alemi, F., 2018. Transport Policy in the Era of Ridehailing and Other Disruptive Transportation Technologies. In: Advances in Transport Policy and Planning. Elsevier, 119–144. Circella, G., Lee, Y., and Alemi, F., 2019. Exploring the Relationships Among Travel Multimodality, Driving Behavior, Use of Ridehailing and Energy Consumption. National Center for Sustainable Transportation. Clewlow, R. and Mishra, G.S., 2017. Shared Mobility: Current Adoption, Use, and Potential Impacts on Travel Behavior. Presented at the Transportation Research Board 96th Annual MeetingTransportation Research Board 2018, Washington DC, USA. Conway, M., Salon, D., and King, D., 2018. Trends in Taxi Use and the Advent of Ridehailing, 1995–2017: Evidence from the US National Household Travel Survey. Urban Science, 2 (3), 79. Currie, G., 2018. Lies, Damned Lies, AVs, Shared Mobility, and Urban Transit Futures. Journal of Public Transportation, 21 (1), 19–30. De Paula, P.C.B. and Zanatta, R.A.F., 2018. The Uber problem in São Paulo: challenges to experimental urban governance. Presented at the 5th International & Comparative Urban Law Conference, São Paulo. Dediu, H., 2019. The Five Categories of Micromobility [online]. Available from: https://micromobility.io/blog/2019/3/20/the-five-categories-of-micromobility. Dias, F.F., Lavieri, P.S., Kim, T., and Bhat, C.R., 2019. Fusing Multiple Sources of Data to Understand Ride- Hailing Use. Presented at the 2019 Transportation Research Board (TRB) Annual Meeting, Washington, D.C., USA. Dowling, R. and Kent, J., 2015. Practice and public–private partnerships in sustainable transport governance: The case of car sharing in Sydney, Australia. Transport Policy, 40, 58–64. Feigon, S. and Murphy, C., 2016. Shared Mobility and the Transformation of Public Transit. Washington, D.C.: Transportation Research Board. Ferguson, E., 1997. The rise and fall of the American carpool: 1970–1990. Transportation, 24 (4), 349–376. Furuhata, M., Dessouky, M., Ordóñez, F., Brunet, M.-E., Wang, X., and Koenig, S., 2013. Ridesharing: The state-of-the-art and future directions. Transportation Research Part B: Methodological, 57, 28–46. Giuliano, G., 1992. Transportation Demand Management: Promise or Panacea? Journal of the American Planning Association, 58 (3), 327–335. Golightly, D., Houghton, R., Hughes, N., and Sharples, S., 2019. Future of Mobility: Evidence Review - Human Factors in Exclusive and Shared Use in the UK Transport System. UK Government Office for Science. Gu, T., Kim, I., and Currie, G., 2019. To be or not to be dockless: Empirical analysis of dockless bikeshare development in China. Transportation Research Part A: Policy and Practice, 119, 122–147. Hardy, A., Hyslop, S., Booth, K., Robards, B., Aryal, J., Gretzel, U., and Eccleston, R., 2017. Tracking tourists’ travel with smartphone-based GPS technology: a methodological discussion. Information Technology & Tourism, 17 (3), 255–274. Heesch, K.C., James, B., Washington, T.L., Zuniga, K., and Burke, M., 2016. Evaluation of the Veloway 1: A natural experiment of new bicycle infrastructure in Brisbane, Australia. Journal of Transport & Health, 3 (3), 366–376. Henao, A. and Marshall, W.E., 2018. The impact of ride-hailing on vehicle miles traveled. Transportation. Hensher, D.A., 1998. The imbalance between car and public transport use in urban Australia: why does it exist? Transport Policy, 5 (4), 193–204. 13

SOAC 2019

Hopkins, D., 2016. Can environmental awareness explain declining preference for car-based mobility amongst generation Y? A qualitative examination of learn to drive behaviours. Transportation Research Part A: Policy and Practice, 94, 149–163. Horowitz, A. and Sheth, J., 1976. Ride sharing to work: A pyschosocial analysis. Chicago, IL, USA: College of Connnerce and Business Administration University of Illinois at Urbana-Champaign. Kaufman, B., Galford, G., and Burke, M., 2019. Optimizing mobile device GPS data collection to capture long distance travel. Presented at the World Conference on Transport Research 2019, Mumbai, India. Kaufman, B. and Leung, A., 2019. We subscribe to movies and music, why not transport? [online]. The Conversation. Available from: http://theconversation.com/we-subscribe-to-movies-and-music-why-not- transport-119538 [Accessed 14 Oct 2019]. Korpilo, S., Virtanen, T., and Lehvävirta, S., 2017. Smartphone GPS tracking—Inexpensive and efficient data collection on recreational movement. Landscape and Urban Planning, 157, 608–617. Kwan, M.-P., 2007. Mobile Communications, Social Networks, and Urban Travel: Hypertext as a New Metaphor for Conceptualizing Spatial Interaction . The Professional Geographer, 59 (4), 434–446. Li, Z., Hong, Y., and Zhang, Z., 2016. Do Ride-Sharing Services Affect Traffic Congestion? An Empirical Study of Uber Entry. SSRN Electronic Journal. ∗ Lovett, M.J. and Peres, R., 2018. Mobile diaries – Benchmark against metered measurements: An empirical investigation. International Journal of Research in Marketing, 35 (2), 224–241. Marchetti, C., 1994. Anthropological invariants in travel behavior. Technological Forecasting and Social Change, 47 (1), 75–88. Mees, P., O’Connell, G., and Stone, J., 2008. Travel to Work in Australian Capital Cities, 1976–2006. Urban Policy and Research, 26 (3), 363–378. Mobility Pricing Independent Commission, 2018. Metro Vancouver mobility pricing study - Full report. Vancouver: Mobility Pricing Independent Commission. Münzel, K., Boon, W., Frenken, K., Blomme, J., and van der Linden, D., 2019. Explaining carsharing supply across Western European cities. International Journal of Sustainable Transportation, 1–12. Nykvist, B. and Whitmarsh, L., 2008. A multi-level analysis of sustainable mobility transitions: Niche development in the UK and Sweden. Technological Forecasting and Social Change, 75 (9), 1373–1387. Punel, A., Ermagun, A., and Stathopoulos, A., 2018. Studying determinants of crowd-shipping use. Travel Behaviour and Society, 12, 30–40. Rayle, L., Dai, D., Chan, N., Cervero, R., and Shaheen, S., 2016. Just a better taxi? A survey-based comparison of taxis, transit, and ridesourcing services in San Francisco. Transport Policy, 45, 168–178. Saadah, K., Yasmine, S.E., and Mubah, A.S., 2017. Digital collaborative consumption and social issues: The clash of taxi and Uber driver in Surabaya and Taipei. Masyarakat, Kebudayaan dan Politik, 30 (4), 333. Saravanan Sundara Sakaran, Harifah Mohd Noor, and Oliver Valentine Eboy, 2018. Socioeconomic Factors that affect Usage of Grabcar Services in Kota Kinabalu City, Sabah. Malaysian Journal of Business and Economics, 5 (2), 65 – 77. Scaraboto, D., 2015. Selling, Sharing, and Everything In Between: The Hybrid Economies of Collaborative Networks. Journal of Consumer Research, 42 (1), 152–176. Shaheen, S., Cohen, A., and Zohdy, I., 2016. Shared Mobility Current Practices and Guiding Principles. Washington DC, USA: US Federal Highway Administration. Shared-Use Mobility Center, 2019. What is Shared Mobility? [online]. Available from: https://sharedusemobilitycenter.org/what-is-shared-mobility/. Smith, C.S., 2017. Canadian town Innisfail hires Uber as its transport system, for ‘better value’ [online]. Australian Financial Review. Available from: https://www.afr.com/technology/canadian-town-innisfail- hires-uber-as-its-transport-system-for-better-value-20170517-gw6g1g [Accessed 14 Oct 2019]. Sprei, F., 2018. Disrupting mobility. Energy Research & Social Science, 37, 238–242. Standing, C., Standing, S., and Biermann, S., 2018. The implications of the sharing economy for transport. Transport Reviews, 1–17. Sun, Y. and Mobasheri, A., 2017. Utilizing Crowdsourced Data for Studies of Cycling and Air Pollution Exposure: A Case Study Using Strava Data. International Journal of Environmental Research and Public Health, 14 (3), 274. Teal, R.F., 1987. Carpooling: Who, how and why. Transportation Research Part A: General, 21 (3), 203– 214. Translink, 2019. TransLink Smart Ticketing Project [online]. translink.com.au. Available from: https://translink.com.au/about-translink/projects-and-initiatives/smartticketing [Accessed 14 Oct 2019].

SOAC 2019

Yim, L.P., 2019. Understanding Supply & Demand in Ride-hailing Through the Lens of Grab Data [online]. Medium. Available from: https://towardsdatascience.com/understanding-supply-demand-in-ride-hailing- through-the-lens-of-grab-data-23e24224547f [Accessed 14 Oct 2019]. Young, M. and Farber, S., 2019. The who, why, and when of Uber and other ride-hailing trips: An examination of a large sample household travel survey. Transportation Research Part A: Policy and Practice, 119, 383–392.